I am rather new to the .SE sites (started with math.SE and then expanded to CV) but during these last three months I had some time on my hands and I have answered a fair number of questions (93 in math.SE and 63 in CV), which can constitute an indicative, if not representative ...sample. This sample tells me that my answers tend to be long, and certainly "against" .SE guidelines that advise for rather short and to-the-point answers.

After some reflection I think this is possibly due to three reasons -two of them are personality traits, the other one may be defensible: a) It is the teacher in me, b) "I like to hear the sound of my own voice" (I read this as an offered explanation for long answers in some math.SE post) and c) The question touches a more general issue that I believe is worth elaborating on in order to put the specific question in perspective, something that I know from off-line experience that makes its answer more useful to the person who asks the question (deeper and more comprehensive understanding).

A recent example is this answer of mine: the OP asked about the meaning of subscripts attached to the symbol of the Expectations operator. Since this phenomenon appears when more random variables are involved, I decided essentially to also answer "another question" -"when more r.v.'s are involved and there is no subscript in the expectations operator, what do we do?". I believe that in this way my answer stands as a more complete treatment of this aspect of scientific notation, it permits the OP and any other reader to understand that the more general question is "when more than one r.v. then what?" -and to have an answer on this last question (which being more general is bound to pop up in more circumstances). But this approach doubled the length of my answer, and strictly speaking, half of the answer was, if not off-topic, at least off-question, as the latter was actually asked.

I think it would be valuable to hear from more experienced members of the community about the matter. As I said, most probably there are "rather questionable" personality traits that are pushing me towards this answering approach, but I honestly believe that it also has something to offer to the readers.

I would really appreciate any thoughts on that.

  • 7
    $\begingroup$ In addition to the replies you receive here, consider an empirical evaluation: take a look at the highest-rated answers on our site and study their characteristics. (You will find some extremely long answers in this collection, by the way.) $\endgroup$
    – whuber Mod
    Commented Oct 13, 2013 at 13:22
  • $\begingroup$ @whuber I like this totally scientific approach, thanks Whuber, I certainly will. $\endgroup$ Commented Oct 13, 2013 at 13:27

4 Answers 4


I'll note that the post you point to is

(i) not all that long

(ii) already well upvoted (including by me; indeed given the active voters and the active meta participants have a big overlap, that's probably true of several people participating in the discussion)

so I think you can take the broader community response as positive already.

But for my two cents

  1. I think long answers (longer than that one) are usually fine. Answers with a pedagogical slant are fine; people are often looking for expositions that will convey understanding, not just an 'answer'. My favourite answers here that I have seen from others are slightly longer ones that undertake to convey understanding. That's not to say that it's not possible to convey understanding succinctly - I've seen great short answers - but that people understand things differently. Sometimes a little more discussion, or a diagram, or an outline of how a result comes about, or some general explanation to help with intuition helps people that won't see it from the shortest answers.

  2. Questions and answers are supposed to help future readers, not just the OP - so making your responses more general than a direct answer to the question is often a good thing. There are sometimes important points and fascinating ideas that are closely related to the question, but that the OP hasn't the knowledge to even know to ask.

Added in edit: your question asks 'should I?' ... but outside general guidance I don't think it's up to us to tell you how best to answer a question outside of what the rules cover (e.g. avoid 'one sentence' answers, avoid link-only answers, and so on) -- one good aspect of CV is the different approaches people bring. There are often several good contributions, some short, some long.

  • 1
    $\begingroup$ +1 for point 2. It is exactly how I think about Q&A activity. $\endgroup$ Commented Oct 14, 2013 at 11:51
  • 3
    $\begingroup$ +1. Perhaps too obvious to mention, but short answers are sometimes the best busy people can do, especially if a question doesn't appear to allow (or deserve) a long answer. The best questions and the best answers here are questions of some generality and challenge, answered in detail and depth. But there is a long tail of less interesting questions, e.g. those too closely tied to someone's particular data, otherwise too specific, too cryptic, etc. We often ignore them, but a short answer may be posted as, one hopes, of some help, or at least as indicating where misconceptions may lie. $\endgroup$
    – Nick Cox
    Commented Oct 15, 2013 at 15:13

This is not necessarily abusing the SE system when you provide extended answers, as long as they can be justified other than by pure digressions. Answering a related subject in passing might also be a good idea, if you feel that this would be useful for the OP and future readers. (In this particular case, the number of upvotes you got suggests you were not too far off base :)

Once you're done with the take-away message, it is fairly easy to add further considerations in your post: You can highlight your main points in the first part, and then append side notes or related topics in the remaining part. Or as @mbq said, you can build your answer incrementally.

Finally, some facts I have noticed with time:

  • Good answers often go beyond what look like seemingly simple questions (that's why it is also important to comment early on new question, i.e. to help clarify what the OP is really looking for).
  • Good answers often increase interest in the issues raised by the OP.
  • 1
    $\begingroup$ Thanks chl. I think your answer converges to @mbq 's one. Both seem to consider expanded answers worthwhile, as long as they are structured in such a way that they can be useful to a variety of readers - to the grabber in a rush, to the careful taker, to the patient digger. $\endgroup$ Commented Oct 13, 2013 at 13:32

You can use the Merkle-tree-like writing scheme, i.e., start with a shortest possible summary of your answer and then more-less repeat it adding details.

This way your post will be a bit longer, but most readers will be able to skip substantial part of it; on the hand you will be able to contain all the information required by a beginner to get it.

  • $\begingroup$ Give the answer, give the proof/argument that supports the answer, expand on the question... a useful suggestion, thanks! $\endgroup$ Commented Oct 13, 2013 at 11:43

Don't think any of your answers are too long. Only thing I'd add to what others have said here is the caveat that CV isn't the place for lengthy expositions of broad topics that are already well covered elsewhere e.g. textbooks (not that your answers fall into this category).

  • 1
    $\begingroup$ That's a sensible self-discipline check. Indeed an answerer should always be asking him or herself "Am I going into quasi-textbook mode"? $\endgroup$ Commented Oct 14, 2013 at 21:14
  • 1
    $\begingroup$ @Alecos At which point outline plus reference is probably better. But, again, I don't see that problem with your answers. $\endgroup$ Commented Oct 14, 2013 at 21:27
  • $\begingroup$ And we don't usually see this problem in general. Even the occasional gigantic walls of text don't look as bad after they've been formatted properly – that's where really long answers usually go wrong. BTW, there's even a character limit on answers. It's huge, something like 60,000, but I hit it once on CogSci. In retrospect, I was probably exceeding fair use of an article, even though it was public domain! $\endgroup$ Commented Jul 31, 2014 at 17:18

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .